Problems with memory eating software

I’m venting right now. This is meant for all software producers but right now specifically for Topaz. I have been a loyal customer for many years, promoting your products to any photographer I come in contact with and within my club.
My problem is that with each new succession of upgrades, updates and new products, software companies, IE, Topaz, you produce items that require more and more memory. How do you expect us to support your products when it becomes necessary for us to continue to upgrade our computers to accommodate using your products.
Not all of us are professional photographers earning, or trying to earn a living off our photos. A large percentage of us are purely amateur and some of us are retired on fix income.
You need to find a way to produce software that doesn’t consume so much memory.
Right now, three of your products won’t work on my computer that worked before the last release of the products.
My computer is only a few years old. I’m not running out to buy another one so I’ll have to find an alternative software that works for me.




Hi Harry, what three products don’t work?

Can you run one of the AI products or Studio 2 and go to the Help-> Graphics info menu, press copy and then post the info here.

Also let me know if your PC has one or two GPU’s.

Before I began trying to use these I made sure that all other programs were closed.

Gigapixel AI
Jpeg to Raw AI
Sharpen AI

Application & Version: Topaz Studio 2 Version 2.2.0

Operating System: Windows 10 (10.0)

Graphics Hardware: Intel(R) HD Graphics 520

OpenGL Driver: 3.3.0 - Build

CPU RAM: 12152 MB

Video RAM: 1500 MB Total, 1200 MB In Use

Preview Limit: 3466 Pixels

Gigapixel for the Mac seems to have trouble using the Intel Iris Plus integrated graphics GPU for its work, but when switching to CPU use it works well.

Harry, your GPU was probably at the limit of what it could do but I understand your frustration. I think your GPU is on the motherboard and if this is a desktop type computer you can probably add in a new video card that will greatly improve performance. If this is a laptop then you have found out why I never recommend them for photo work. Most people want to get better software and that usually demands better hardware over time. Here is a review of the Intel HD 520 and you can see it is not very good for this type of application. Intel HD 520 (Review) – Graphics of 6th Gen Core U-Series CPUs – Laptop Graphics

1 Like

I hate to say it, but that iGPU is just not going to cut it with the AI SW. That integrated GPU is also where I started and although the early SW versions did run, they were painfully slow. I added an outboard GPU with 4GB memory and the difference is like night & day. The good news is that the cost was only ~$125 so you shouldn’t have to buy a new system assuming your HW is not a laptop.

1 Like

@fotoman I wholeheartedly agree Harry. I stopped getting updates many versions ago. My system can not function with the AI products and to be fair I have never wanted the AI products. I pretty much only use the original version of Topaz Studio. While. apparently on the roadmap for inclusion, to me it seems the functionally offered in the original Studio, such as image layer or other tools such as cloning or perspective altering, have been way-layed by development of AI products. I too am sadden by the direction Topaz has taken.


I’ll look into that hardware. Do you have a brand name?

I agree Shazzy, I’m not all that impressed with the AI products either. I much prefer the originals myself. Particularly DeNoise and Remask over the new products. But, I still like the Topaz Products/


Sapphire Radeon RX570. This is from April-2019.


1 Like

Thanks jgr00ms2 but I just realized I didn’t understand your post. I thought this was a piece of hardware that would plug into my laptop.

That is why I said “assuming this is not a laptop”.

External GPU’s are quite challenging from a HW standpoint due to the massive amounts of data that require to be passed and the HW paths necessary. Thunderbolt interfaces may be able to do it on Mac HW, but I don’t know if it is even feasible on WinTel. I doubt that it would be practical; it would probably be more cost effective to just use a desktop with an outboard GPU video card.

As some of us mentioned previously, laptop HW is not the best way to go for GPU intensive SW for precisely this reason. The flexibility to upgrade HW is severely hamstrung. I have laptops and I sometimes use them for photo editing while traveling, but with a scaled down workflow from my photo desktop ie they don’t really have what it takes HW-wise to duplicate what I use at home on the desktop.

1 Like

AI is expensive in terms of download size and computing resources, and a zillion times slower to run and I don’t see much difference, in some cases the result is very poor. Maybe it’s time to reconsider.

1 Like

I feel that AI has shown that, in the case of image processing, it can give the lowest common denominator as a result.

But I don’t see AI takeing into account deviations from the norm so therefore it is only suitable for what we would call ‘standard’ images using standard settings. By standard settings I mean shooting using Program, Aperture, Tv or full Auto mode or even Manual with one auto parameter such as ISO, aperture or time.

When you deviate from the ‘mean’ immediately you introduce unknown factors into the equation. For example it doesn’t handle high ISO in strong daylight where you want to use a really fast shutter speed to improve a portrait in harsh conditions. It also doesn’t handle properly images where you shoot to expose for shadows or highlights.

So, the question is, is it analyzing the image or is it analyzing the image information?

There are other RAW image processors that already give a different analytical starting point for adjustments such as noise reduction and sharpening based on the camera type and ISO rating.

But, for the majority of images, AI gives a much better result than most people could manager with manual adjustments.

So, in summary, I would like to see full Auto and Full Manual as an alternative. That would remove the reliance on hardware to handle the auto processing.


Another comparison I would raise is between AI Mask and the sky replacement tools that are cropping up in other software.

Using AI Mask for sky replacement is slow and not very accurate, particularly around trees. The sky replacement tool in Luminar 4 is extremely impressive and also lightening fast in comparison. Having this tool within the editing environment, as opposed to a plug-in, is also very useful.

It adds substance to the questions many are asking Topaz about their standalone AI products.

1 Like


As a Luminar 4.1 user, I can say that whilst its AI sky replacement does an excellent job with some photos, it can need some additional help to get an acceptable result with others, and there are some photos where it will not even attempt to do the replacement.

I am not being critical at all of that function. It was designed to work with photos where the original sky falls into a certain category, and it does indeed work extremely well in those situations. Unfortunately though users expectations “because it is AI” can exceed the ability of the software.

My personal view is the main benefit of AI based end user image editing software is that it can substantially reduce the time compared with the manual alternative, and it can also reduce the operator experience needed.

What needs to be accepted though is that AI will not always “get it right”.

1 Like

2 posts were split to a new topic: DeNoise AI v2 running slower